Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method of processing 360-degree virtual reality images, the method comprising: receiving input data for a 2D (two-dimensional) frame, wherein the 2D frame is projected from a 3D (three-dimensional) sphere using a target projection; dividing the 2D frame into multiple blocks; and encoding or decoding said multiple blocks using quantization parameters by restricting a delta quantization parameter to be within a threshold for any two blocks corresponding to two neighboring blocks on the 3D sphere, wherein the delta quantization parameter is restricted to (+/−)x, and wherein x is an integer greater than 0 and smaller than a maximum delta quantization parameter for any two blocks in a whole frame of the 2D frame.
This invention relates to processing 360-degree virtual reality (VR) images, specifically addressing challenges in encoding and decoding 2D frames derived from 3D spherical projections. The method involves receiving a 2D frame generated by projecting a 3D spherical image using a target projection technique. The 2D frame is divided into multiple blocks, which are then encoded or decoded using quantization parameters. A key aspect is restricting the delta quantization parameter—the difference in quantization between neighboring blocks on the 3D sphere—to a predefined threshold. This threshold is defined as an integer value x, where x is greater than 0 but smaller than the maximum allowed delta quantization parameter for any two blocks in the entire frame. By limiting the delta quantization parameter, the method ensures smoother transitions and better visual quality in the reconstructed 360-degree VR content, particularly when neighboring blocks on the sphere are processed. This approach helps mitigate artifacts and inconsistencies that can arise from abrupt changes in quantization levels, improving the overall efficiency and fidelity of VR image processing.
2. The method of claim 1 , wherein the target projection corresponds to Equirectangular Projection (ERP), Cubemap Projection (CMP), Adjusted Cubemap Projection (ACP), Equal-Area Projection (EAP), Octahedron Projection (OHP), Icosahedron Projection (ISP), Segmented Sphere Projection (SSP), Rotated Sphere Projection (RSP), or Cylindrical Projection (CLP).
This invention relates to methods for processing and projecting 360-degree immersive media content, such as panoramic images or videos, onto various projection formats. The core challenge addressed is the efficient and accurate transformation of spherical or 360-degree content into different projection formats while preserving visual quality and minimizing computational overhead. The method involves selecting a target projection format from a predefined set of options, including Equirectangular Projection (ERP), Cubemap Projection (CMP), Adjusted Cubemap Projection (ACP), Equal-Area Projection (EAP), Octahedron Projection (OHP), Icosahedron Projection (ISP), Segmented Sphere Projection (SSP), Rotated Sphere Projection (RSP), or Cylindrical Projection (CLP). Each projection format is optimized for specific use cases, such as reducing distortion, improving rendering efficiency, or supporting different display technologies. The method ensures that the input 360-degree content is accurately mapped to the chosen projection, enabling compatibility with various playback and processing systems. This approach enhances flexibility in immersive media workflows, allowing content creators and distributors to adapt content for different platforms and devices while maintaining high-quality visual output.
3. An apparatus for processing 360-degree virtual reality images, the apparatus comprising one or more electronic devices or processors configured to: receive input data for a 2D (two-dimensional) frame, wherein the 2D frame is projected from a 3D (three-dimensional) sphere using a target projection; divide the 2D frame into multiple blocks; and encode or decode said multiple blocks using quantization parameters by restricting a delta quantization parameter to be within a threshold for any two blocks corresponding to two neighboring blocks on the 3D sphere, wherein the delta quantization parameter is restricted to (+/−)x, and wherein x is an integer greater than 0 and smaller than a maximum delta quantization parameter for any two blocks in a whole frame of the 2D frame.
This invention relates to processing 360-degree virtual reality (VR) images, specifically addressing challenges in encoding and decoding such images while maintaining visual quality and compression efficiency. The apparatus includes one or more electronic devices or processors configured to handle 2D frames derived from 3D spherical projections. The system receives input data for a 2D frame, which is generated by projecting a 3D sphere onto a 2D plane using a target projection method. The 2D frame is then divided into multiple blocks for processing. During encoding or decoding, these blocks are processed using quantization parameters, with a key constraint: the delta quantization parameter (the difference in quantization levels between adjacent blocks) is restricted to a predefined threshold. This threshold is set to a value x, where x is an integer greater than 0 but smaller than the maximum allowed delta quantization parameter for any two blocks in the entire frame. The restriction ensures that neighboring blocks on the 3D sphere, when mapped to the 2D frame, maintain consistent quantization levels, preventing abrupt quality changes and artifacts in the reconstructed VR image. This approach improves compression efficiency while preserving visual coherence in 360-degree VR content.
4. A method of processing 360-degree virtual reality images, the method comprising: receiving input data for a 2D (two-dimensional) frame, wherein the 2D frame is projected from a 3D (three-dimensional) sphere using a target projection; adding one or more guard bands to one or more edges that are discontinuous in the 2D frame but continuous in the 3D sphere, wherein said one or more guard bands are filled with padding data; applying a fade-out process to said one or more guard bands, wherein the fade-out process comprises blending said one or more guard bands with an overlapped region when the overlapped region exists and the fade-out process comprises blending said one or more guard bands with a predefined region when the overlapped region does not exist; and encoding the 2D frame extended with said one or more guard bands.
This invention relates to processing 360-degree virtual reality (VR) images to address discontinuities that arise when projecting a 3D spherical image onto a 2D frame. The problem occurs because spherical projections create seams or edges in the 2D frame that are discontinuous, even though they represent a continuous surface in the 3D sphere. These discontinuities can cause visual artifacts during encoding and playback. The method involves receiving a 2D frame generated from a 3D spherical image using a target projection technique. To mitigate discontinuities, one or more guard bands are added to the edges of the 2D frame where these discontinuities occur. The guard bands are filled with padding data to ensure smooth transitions. A fade-out process is then applied to the guard bands, blending them with an overlapped region if one exists or with a predefined region if no overlap is present. This blending helps reduce visible seams. Finally, the 2D frame, now extended with the processed guard bands, is encoded for storage or transmission. The technique improves visual quality by minimizing artifacts caused by spherical-to-2D projection discontinuities.
5. The method of claim 4 , wherein said one or more guard bands are filled using geometry padding, and wherein the geometry padding extends samples outside said one or edges of the 2D frame using neighboring samples on the 3D sphere.
This invention relates to image processing techniques for handling guard bands in 2D frames derived from 3D spherical data, such as panoramic or immersive video. The problem addressed is the distortion or artifacts that occur at the edges of 2D frames when they are extracted from a 3D spherical representation, particularly when the frame boundaries do not align perfectly with the spherical geometry. These issues can lead to visual discontinuities or loss of information near the edges. The solution involves filling the guard bands—regions near the edges of the 2D frame—using geometry padding. This padding technique extends samples beyond the frame's boundaries by leveraging neighboring samples from the 3D spherical data. Instead of simply repeating or interpolating edge pixels, the method uses the spherical geometry to determine the most appropriate neighboring samples, ensuring seamless transitions and preserving visual quality. This approach is particularly useful in applications like virtual reality, 360-degree video, and other immersive media where maintaining edge integrity is critical. The padding process dynamically adapts to the spherical structure, minimizing artifacts and improving the overall viewing experience.
6. The method of claim 4 , wherein said one or more guard bands are filled by extending boundary samples of said one or more edges.
This invention relates to digital signal processing, specifically methods for handling guard bands in image or signal data to prevent artifacts during processing. The problem addressed is the occurrence of visual or auditory artifacts when processing signals with sharp transitions or edges, such as in image scaling, compression, or audio filtering. These artifacts arise due to insufficient handling of boundary regions between processed and unprocessed areas, often referred to as guard bands. The method involves filling one or more guard bands by extending boundary samples of edges within the signal. This means that when a signal contains distinct edges or transitions, the samples near these edges are replicated or interpolated to fill adjacent guard band regions. This approach ensures smooth transitions and reduces artifacts that would otherwise occur due to abrupt changes in the signal. The technique is particularly useful in applications like image resizing, where maintaining edge integrity is critical, or in audio processing, where edge artifacts can cause distortion. The method may be applied to various types of signals, including but not limited to images, video frames, and audio waveforms. By extending boundary samples into guard bands, the technique maintains signal integrity while allowing efficient processing of the core data. This approach is distinct from other methods that may rely on padding with zeros or fixed values, as it dynamically adapts to the signal's natural edges. The result is improved visual or auditory quality with minimal computational overhead.
7. The method of claim 4 , wherein said one or more guard bands are filled with duplicated samples from respective edge areas of said one or more edges.
This invention relates to digital signal processing, specifically methods for handling guard bands in data transmission or storage systems. Guard bands are regions of data that separate different segments or channels to prevent interference. The problem addressed is ensuring data integrity and minimizing errors when processing signals with guard bands, particularly in scenarios where edge effects or boundary conditions may cause distortions. The method involves filling one or more guard bands with duplicated samples from the edge areas of the adjacent data segments. By replicating edge samples into the guard band regions, the system maintains continuity and reduces abrupt transitions that could lead to signal degradation. This approach is particularly useful in applications like digital communications, where guard bands separate frequency channels, or in data storage systems where guard bands isolate different data blocks. The duplicated samples are taken from the immediate edge areas of the data segments bordering the guard band. This ensures that the guard band content is consistent with the adjacent data, minimizing phase or amplitude discontinuities. The method can be applied to various types of signals, including time-domain or frequency-domain representations, depending on the system requirements. The technique helps maintain signal quality, reduce interference, and improve overall system performance by mitigating edge effects in guard band regions.
8. The method of claim 4 , wherein the target projection corresponds to Equirectangular Projection (ERP), Cubemap Projection (CMP), Adjusted Cubemap Projection (ACP), Equal-Area Projection (EAP), Octahedron Projection (OHP), Icosahedron Projection (ISP), Segmented Sphere Projection (SSP), Rotated Sphere Projection (RSP), or Cylindrical Projection (CLP).
This invention relates to methods for processing and projecting three-dimensional (3D) data, particularly for converting 3D data into various projection formats. The problem addressed is the need for flexible and efficient projection techniques to represent 3D data in different formats suitable for various applications, such as virtual reality, computer graphics, and data visualization. The method involves converting 3D data into a target projection format, where the target projection can be one of several standardized or custom projection types. These include Equirectangular Projection (ERP), which maps a sphere onto a 2D rectangle; Cubemap Projection (CMP), which divides the 3D space into six square faces; Adjusted Cubemap Projection (ACP), a modified version of CMP; Equal-Area Projection (EAP), which preserves area relationships; Octahedron Projection (OHP), which uses an octahedron to map 3D data; Icosahedron Projection (ISP), which uses an icosahedron; Segmented Sphere Projection (SSP), which divides a sphere into segments; Rotated Sphere Projection (RSP), which rotates the sphere before projection; and Cylindrical Projection (CLP), which maps data onto a cylinder. The method ensures that 3D data can be accurately transformed into any of these projection formats, enabling compatibility with different rendering systems and applications. This flexibility allows for optimized performance and visualization quality depending on the specific requirements of the target platform or use case. The invention improves upon existing techniques by providing a unified approach to projection conversion, reducing the need for multiple specialized algorithms.
9. A method of processing 360-degree virtual reality images, the method comprising: receiving coded data for an extended 2D (two-dimensional) frame including an encoded 2D frame with one or more encoded faded guard bands, wherein the encoded 2D frame is projected from a 3D (three-dimensional) sphere using a target projection, wherein said one or more encoded faded guard bands are based on a blending of one or more guard bands with an overlapped region when the overlapped region exists and a blending of one or more guard bands with a predefined region when the overlapped region does not exist; decoding the coded data into a decoded extended 2D frame including a decoded 2D frame with one or more decoded faded guard bands; and deriving a reconstructed 2D frame from the decoded extended 2D frame.
This invention relates to processing 360-degree virtual reality (VR) images, specifically addressing the challenge of efficiently encoding and decoding spherical 3D content into a 2D format for display. The method involves receiving coded data for an extended 2D frame, which includes an encoded 2D frame with one or more encoded faded guard bands. The encoded 2D frame is generated by projecting a 3D spherical image onto a 2D plane using a target projection method. The guard bands are regions at the edges of the 2D frame that help mitigate visual artifacts when the content is stitched or blended. These guard bands are encoded with a fading effect, which is determined by blending the guard bands with either an overlapped region (if present) or a predefined region (if no overlap exists). The coded data is then decoded into a decoded extended 2D frame, which retains the original 2D frame along with the faded guard bands. Finally, a reconstructed 2D frame is derived from the decoded extended 2D frame, ensuring smooth transitions and minimizing distortion in the final VR image. This approach improves the quality and efficiency of 360-degree VR image processing by optimizing the handling of edge regions during encoding and decoding.
10. The method of claim 9 , wherein the reconstructed 2D frame is generated from the decoded extended 2D frame by cropping said one or more decoded faded guard bands.
Video compression techniques often introduce guard bands around decoded frames to prevent artifacts, but these bands reduce image quality and resolution. This invention addresses the problem by reconstructing a high-quality 2D frame from a decoded extended 2D frame that includes faded guard bands. The method involves decoding an extended 2D frame, which contains the original image data surrounded by faded guard bands. These guard bands are then cropped to remove the faded regions, resulting in a reconstructed 2D frame that matches the original dimensions and quality. The fading of the guard bands ensures smooth transitions at the edges, minimizing visible artifacts. This approach improves image fidelity by eliminating unnecessary guard band regions while maintaining visual quality. The method is particularly useful in video encoding and decoding systems where guard bands are used to prevent edge artifacts during compression. By dynamically adjusting the guard band removal process, the system ensures optimal image reconstruction without compromising performance. This technique enhances the efficiency of video processing pipelines by reducing unnecessary data while preserving image integrity.
11. The method of claim 9 , wherein the target projection corresponds to Equirectangular Projection (ERP), Cubemap Projection (CMP), Adjusted Cubemap Projection (ACP), Equal-Area Projection (EAP), Octahedron Projection (OHP), Icosahedron Projection (ISP), Segmented Sphere Projection (SSP), Rotated Sphere Projection (RSP), or Cylindrical Projection (CLP).
This invention relates to methods for processing and projecting three-dimensional (3D) spatial data, particularly for converting between different projection formats used in virtual reality (VR), augmented reality (AR), and 3D visualization applications. The problem addressed is the need for efficient and accurate conversion between various projection formats to ensure compatibility across different systems and platforms. The method involves transforming 3D spatial data from one projection format to another, where the target projection can be any of several standardized formats, including Equirectangular Projection (ERP), Cubemap Projection (CMP), Adjusted Cubemap Projection (ACP), Equal-Area Projection (EAP), Octahedron Projection (OHP), Icosahedron Projection (ISP), Segmented Sphere Projection (SSP), Rotated Sphere Projection (RSP), or Cylindrical Projection (CLP). Each of these projections has distinct characteristics suited for different applications, such as minimizing distortion, optimizing storage, or improving rendering performance. The method ensures that the conversion process preserves spatial accuracy and visual quality, allowing seamless integration of 3D data across various devices and software environments. This is particularly useful in applications requiring real-time rendering, such as VR headsets, 3D mapping, and immersive media. The invention provides a flexible solution for developers and engineers working with 3D spatial data, enabling efficient adaptation to different projection requirements without manual intervention.
Unknown
April 7, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.